On Sun, 16 Oct 2016 14:56:14 +0000 "Laurent Bercot" <ska-skaw...@skarnet.org> wrote:
> >So, please enlighten me if I'm missing a better solution, of course, > >but in case I'm not, I was wondering how you'd feel about adding a > >new option to execlineb, e.g. -s, that would require a number as arg > >(much like -S) but would work as such: > Still considering your -s idea, which should be easy enough to > implement once I remember how execlineb does things :) but in the > meantime, wouldn't the "runblock" command help you? Something like > > #!/command/execlineb > importas FIRSTARG 1 > # work with $FIRSTARG > runblock -n 1 -r -- 0 Well, that still involves the environment, which I'd like to skip altogether, but it does seem a bit "easier" since the runblock call replaces shift+elgetpositionals+emptyenv; However, this gives me "odd" results. Consider your script above as ./helper : $ ./helper foo bar baz runblock: fatal: unable to exec bar: No such file or directory Yep, makes sense. Now trying this: $ ./helper foo bar Nothing. It ran the helper, then didn't (try to) exec into bar. Now let's try this: $ ./helper foo and it segfaults! (So does "./helper" btw) So maybe "./helper foo bar" should actually work, but doesn't because of the same bug that causes segfault?